GPT-5.6 Luna · 官方评测披露
GPT-5.6 Luna 当前收录 24 项测试的官方记录。已完成列明材料的核查。
模型发布 · GPT Luna
官方材料与核查范围
The retained GPT-5.6 article edition, eight direct-effort interactive charts for this model, and the current 82-page system card were fully reviewed. Chart-state disclosure dates and latest upstream page revisions are not stated; no launch timing is inferred from the page date label. 核对时间 。
当前资料不足以还原首发。 没有可核实的首发材料版本快照。
- GPT-5.6: Frontier intelligence that scales with your ambition | OpenAI · 主文 · Official GPT-5.6 article page cache; page date label 2026-07-09, with later price updates dated 2026-07-30 and 2026-08-21. The 2026-11-09 browser capture includes the article tables and interactive chart labels; the chart-state disclosure date is not stated.公布 未确认 · 采集 · 已核查 The retained 2026-11-09 official page capture and all source tables were reviewed. This batch checked all 8 model-specific interactive charts that expose an explicit effort and score; 45 exact GPT-5.6 Luna points were read from their point labels. The page's latest upstream revision remains unknown.
- GPT-5.6 System Card · 附件 · Official page title is GPT-5.6 System Card. Current HTML snapshot; publication date of this page version is not stated.公布 未确认 · 采集 · 已核查 Current GPT-5.6 system-card HTML and all 82 PDF pages were read in full and visually checked page by page; model-specific ordinary benchmark rows and figures were checked. Safety and capability-risk studies were read but excluded from regular results.
配置
- GPT-5.6 Luna · 档位未注明 · 主文表格未逐行标明本型号推理档位;不套用 Sol 的 effort,也不以价格推断档位。 未知项 推理档位、部分评测的完整 harness 与采样设置
- GPT-5.6 Luna (low) · GPT-5.6 Luna; reasoning effort Low as directly labeled on the official chart. Low未知项 Per-benchmark sampling and chart harness settings are not stated.
- GPT-5.6 Luna (medium) · GPT-5.6 Luna; reasoning effort Medium as directly labeled on the official chart. Medium未知项 Per-benchmark sampling and chart harness settings are not stated.
- GPT-5.6 Luna (high) · GPT-5.6 Luna; reasoning effort High as directly labeled on the official chart. High未知项 Per-benchmark sampling and chart harness settings are not stated.
- GPT-5.6 Luna (xhigh) · GPT-5.6 Luna; reasoning effort Xhigh as directly labeled on the official chart. Xhigh未知项 Per-benchmark sampling and chart harness settings are not stated.
- GPT-5.6 Luna (max) · GPT-5.6 Luna; reasoning effort Max as directly labeled on the official chart. Max未知项 Per-benchmark sampling and chart harness settings are not stated.
- GPT-5.6 Luna (none) · GPT-5.6 Luna; reasoning effort None as directly labeled on the official chart. None未知项 Per-benchmark sampling and chart harness settings are not stated.
直接前代与后续发布
当前收录未确认直接前代。 官方关系依据
We’re launching the GPT‑5.6 family of models for general availability following our limited preview: our new flagship, Sol, alongside Terra, a balanced model for everyday work, and Luna, our most cost-efficient model.
测试记录 · 24
- Agents’ Last Exam · 55 professional fields; exact task version unspecified本次发布有 5 条成绩记录,含 5 条精确值(可含历史记录)
- GDPval-AA · v2本次发布有 7 条成绩记录,含 7 条精确值(可含历史记录)
- Big Finance Bench · Version and task protocol unspecified本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- Artificial Analysis Intelligence Index · v4.1本次发布有 7 条成绩记录,含 7 条精确值(可含历史记录)
- Artificial Analysis Coding Agent Index · v1.1本次发布有 6 条成绩记录,含 6 条精确值(可含历史记录)
- SWE-Bench Pro · Subset/version unspecified in source本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- DeepSWE · v1.1本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- Terminal-Bench · 2.1; source metric reported as score percentage本次发布有 6 条成绩记录,含 6 条精确值(可含历史记录)
- GeneBench Pro · Version unspecified in source本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- LifeSciBench · Version unspecified in source本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- HealthBench Professional · 525 physician-authored conversations本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- OSWorld · 2.0; subset and scoring details unspecified本次发布有 7 条成绩记录,含 7 条精确值(可含历史记录)
- BrowseComp · 1,266 题;单代理与多代理按协议区分本次发布有 6 条成绩记录,含 6 条精确值(可含历史记录)
- BenchCAD · Task version and scoring details unspecified本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- BenchCAD · Task version unspecified; Python tool protocol本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- MMMU-Pro · Standard and Vision formats本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- MMMU-Pro · Tool-enabled protocol; exact version unspecified (gpt56-mmmu-pro-with-tools)本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- GDP.pdf · 100 professional PDFs and prompts本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- GPQA Diamond · 198-question Diamond subset本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- FrontierMath · Tier 1–3 v2本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- FrontierMath · Tier 4 v2本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- Toolathlon · Version, subset, and metric unspecified in the 2026-07-09 article本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- ARC-AGI-3 · Version/subset unspecified in source本次发布有 1 条成绩记录,含 1 条精确值(可含历史记录)
- AutomationBench · Zapier 私有 early-access set;版本未注明本次发布有 6 条成绩记录,含 6 条精确值(可含历史记录)